AN-1177: LVDS and M-LVDS Circuit Implementation Guide (Rev. 0) PDF

AN-1177
APPLICATION NOTE
One Technology Way • P.O. Box 9106 • Norwood, MA 02062-9106, U.S.A. • Tel: 781.329.4700 • Fax: 781.461.3113 • www.analog.com
LVDS and M-LVDS Circuit Implementation Guide
by Dr. Conal Watterson
INTRODUCTION
LVDS/M-LVDS APPLICATION CONSIDERATIONS
Low voltage differential signaling (LVDS) is a standard for
communicating at high speed in point-to-point applications.
Multipoint LVDS (M-LVDS) is a similar standard for multipoint applications. Both LVDS and M-LVDS use differential
signaling, a two-wire communication method where receivers
detect data based on the voltage difference between two
complementary electrical signals. This greatly improves noise
immunity and minimizes emissions.
This application note considers the following aspects
concerning LVDS/M-LVDS circuit implementation:
LVDS is a lower power alternative to emitter-coupled logic
(ECL) or positive emitter-coupled logic (PECL).The primary
standard for LVDS is TIA/EIA-644. An alternative standard
sometimes used for LVDS is IEEE 1596.3—SCI, scalable
coherent interface. LVDS has been widely adopted for highspeed backplane, cabled, and board-to-board data transmission
and clock distribution, as well as communication links within a
single PCB.
Advantages of LVDS include
•
•
•
•
•
Communication at speeds of up to 1 Gbps or more
Reduced electromagnetic emissions
Increased immunity to noise
Low power operation
Common-mode range allowing differences of up to ±1 V
in ground offset
Bus types and topologies
Clock distribution applications
Characteristics of LVDS/M-LVDS signaling
Termination and PCB layout
Jitter and skew
Data encoding and synchronization
Isolation
WHY USE LVDS OR M-LVDS?
LVDS and M-LVDS are compared to other multipoint and pointto-point protocols in Figure 1. Both standards have low power
requirements. LVDS and M-LVDS are characterized by differential
signaling with a low differential voltage swing. M-LVDS specifies
an increased differential output voltage compared to LVDS in order
to allow for the increased load from a multipoint bus.
Both protocols are designed for high-speed communication.
Typical applications utilize PCB traces or short wired/backplane
links. The common mode range of LVDS is designed for these
applications. M-LVDS has an extended common mode range
compared to LVDS to allow for the additional noise in a multipoint
topology.
M-LVDS
The standard TIA/EIA-899 for multipoint low voltage differential signaling (M-LVDS) extends LVDS to address multipoint
applications. M-LVDS allows higher speed communication
links than TIA/EIA-485 (RS-485) or controller area network
(CAN) with lower power. See the References section for a list of
the standards referred to in this application note.
MULTIPOINT
M-LVDS
LOW POWER, HIGH SPEED
MEDIUM DISTANCES (MAX. 20m TO 40m)
TYP. DATA RATE: 100Mbps, 200Mbps
RS-485
LONG DISTANCES (>1km)
TYP. MAX. DATA RATE: 16Mbps
CAN
ROBUST PROTOCOL
MEDIUM DISTANCES (MAX. 40m)
MAX. DATA RATE: 1Mbps
POINT-TO-POINT
Additional features of M-LVDS over LVDS include
•
•
•
•
Increased driver output strength
Controlled transition times
Extended common-mode range
Option of failsafe receivers for bus idle condition
LVDS
LOW POWER, HIGH SPEED
SHORT DISTANCES (MAX. 5m TO 10m)
MAX. DATA RATE: >1Gbps
PECL
HIGH SPEED
SHORT DISTANCES
MAX. DATA RATE: ~3Gbps
Figure 1. Comparison of Communication Standards
Rev. 0 | Page 1 of 12
11236-001
LVDS
•
•
•
•
•
•
•
AN-1177
Application Note
TABLE OF CONTENTS
Introduction ...................................................................................... 1
Definitions and Output Levels ....................................................5
LVDS/M-LVDS Application Considerations ................................ 1
Receiver Thresholds ......................................................................5
Why use LVDS or M-LVDS? ........................................................... 1
Transmission Distance..................................................................6
Revision History ............................................................................... 2
Termination and PCB Layout ..........................................................7
Bus Types and Topologies ............................................................... 3
Controlled Impedances ................................................................7
Point-to-Point ............................................................................... 3
Jitter, Skew, Data Encoding, and Synchronization ........................8
Multi-Drop .................................................................................... 3
What is Jitter? .................................................................................8
Multipoint ...................................................................................... 3
What is Skew? ................................................................................8
Clock Distribution Applications ..................................................... 4
Data Encoding and Synchronization ..........................................9
Multi-Drop Clock Distribution .................................................. 4
Isolation ........................................................................................... 10
Point-to-Point Clock Distribution ............................................. 4
References ........................................................................................ 11
Clock Distribution Using M-LVDS ........................................... 4
Related Links ............................................................................... 11
Differential Signalling and LVDS/M-LVDS .................................. 5
REVISION HISTORY
3/13—Revision 0: Initial Version
Rev. 0 | Page 2 of 12
Application Note
AN-1177
BUS TYPES AND TOPOLOGIES
MULTIPOINT
Standard TIA/EIA-644 LVDS devices allow low power, high
speed communication. The advantages of LVDS can also
be applied to multipoint applications by using TIA/EIA-899
devices. Bus topology is one of the main factors relating to
which LVDS or M-LVDS devices are used in an application.
POINT-TO-POINT
Point-to-point bus topologies consist of a single driver and single
receiver connected together using one pair of wires or traces.
Figure 2 demonstrates a typical configuration, where the receiving
end of the link has a termination resistor. This is the most common
application for LVDS devices. Multiple pairs of wires or traces
can be used to create additional channels of communication and
increase total bandwidth between two points.
DOUT+
In networks where multiple devices can either send or receive,
a multipoint bus topology may be used. M-LVDS is designed
for such multi-point applications, allowing up to 32 nodes to
be connected to a single bus. There are two types of multipoint
buses, half-duplex and full duplex, shown in Figure 4 and
Figure 5, respectively. In a half-duplex bus, two wires are used
such that one device may transmit, and the other devices can
receive. In a full-duplex bus, four wires are used, allowing one
node to concurrently transmit back to another transmitting
node (that is, slave devices responding as broadcast commands
are sent by the master to all nodes).
A
RO
RIN+
Figure 2. LVDS Point-to-Point Link
DI
Tx
4
0
4
0
Rx
0
4
0
4
RO
RT
DI
Z
DI
RO
DI
RO
Figure 5. M-LVDS Full-Duplex Bus
Another factor to be considered in multipoint buses is the bus
idle condition. When no device is transmitting, the differential
voltage on a terminated bus will be close to 0 V. This means that
for a standard receiver with symmetrical input thresholds, the
receiver output will be undefined. This corresponds to the
Type 1 M-LVDS receivers with an input threshold of ±50 mV.
In order to provide a guaranteed receiver output state (output
low) in the bus idle condition, Type 2 M-LVDS receivers have
an offset receiver input threshold of +50 mV to +150 mV.
Table 2. M-LVDS Transceivers
RT
RIN–
ROUT
Figure 3. LVDS Multi-Drop Bus
11236-003
LVDS
RECEIVERS
ROUT
RT
B
Y
A
B
A single driver can be connected to multiple receivers using
a multi-drop bus topology as shown in Figure 3. LVDS is
designed for point-to-point applications and so in a multi-drop
configuration, the number of receivers that can be connected
and the signaling distance can be limited. M-LVDS can be used
in a multi-drop topology to drive up to 32 nodes across longer
distances compared to LVDS.
DOUT–
RT
MLVDS
TRANSCEIVERS
MULTI-DROP
RIN+
RT
Z
RO
M-LVDS can also be used in a point-to-point topology, where the
same transceiver device is used for the driver circuit (with receiver
disabled) and the receiving circuit (with driver disabled).
DOUT+
A
DI
Part No.
ADN4665
ADN4666
ADN4667
ADN4668
DIN
RO
Figure 4. M-LVDS Half-Duplex Bus
Y
Table 1. LVDS Drivers and Receivers
Rx
0
1
0
2
DI
MLVDS
TRANSCEIVERS
Analog Devices, Inc., has a portfolio of LVDS drivers and receivers
for one, two or four LVDS channels as shown in Table 1. Unused
outputs should be left open circuit.
Tx
1
0
2
0
B
11236-005
RIN–
B
RO
11236-004
DOUT–
LVDS
RECEIVER
11236-002
RT
LVDS
DRIVER
LVDS
DRIVER
RT
RT
ROUT
DIN
Part No.
ADN4661
ADN4662
ADN4663
ADN4664
A
DI
Part No.
ADN4690E
ADN4691E
ADN4692E
ADN4693E
ADN4694E
ADN4695E
ADN4696E
ADN4697E
Rev. 0 | Page 3 of 12
Rx Type
1
1
1
1
2
2
2
2
Duplex
Half
Half
Full
Full
Half
Full
Half
Full
Data Rate
100
200
100
200
100
100
200
200
AN-1177
Application Note
CLOCK DISTRIBUTION APPLICATIONS
Differential signaling, such as LVDS, is a good choice for
distributing clock signals around a circuit board. In addition to
the benefits of the common-mode noise immunity of LVDS, a
particular advantage for clock distribution applications is that
radiated emissions are reduced due to the coupling between the
two opposing signals.
CK
SI
11-BIT CONTROL
REGISTER
MUX
1
0
CLK
Q8
CLK0
CLK0
0
CLK1
CLK1
1
Q7
MUX
Q7
Q6
Q6
Q5
CLK
CLOCK
SOURCE
RT
Q5
Q4
RIN–
DOUT–
NODE 9
Q9
Q8
RIN+
DOUT+
10 LVDS POINTTO-POINT LINKS
Q9
In many applications, multiple nodes in a circuit may depend
on a single clock source. A simple approach to distributing
a single clock source to multiple nodes using LVDS, is to use
a multi-drop bus topology as shown in Figure 6. The LVDS
outputs of a clock source are connected to a pair of signal traces
that have short stubs to the various nodes relying on the clock.
LVDS
CLOCK
SOURCE
12-BIT COUNTER
EN
CLOCK
SOURCE
MULTI-DROP CLOCK DISTRIBUTION
11-BIT SHIFT
REGISTER
Q4
LVDS
CLOCK INPUTS
Q3
Q3
Q2
Q2
Figure 6. Multi-Drop LVDS Clock Distribution
Q1
Q1
The disadvantages of this approach are that the number of
nodes that can be connected is limited and stubs contribute to
degradation of the signal integrity (that is, adding jitter). Stub
lengths and impedances must be carefully controlled.
POINT-TO-POINT CLOCK DISTRIBUTION
A single clock source can be connected to a single node
requiring an LVDS clock input using a point-to-point link.
This can be extended to supply multiple nodes by means of
an LVDS buffer acting as a fan-out device. This separate
component receives the LVDS clock output from the clock
source, and in turn provides this clock signal to multiple LVDS
drivers in the device to drive multiple point-to-point links to
receiving nodes. The advantage of this approach is that timing
on the clock signal can remain unaffected by stubs.
An example of such a device is the ADN4670 clock distribution
buffer. This allows one of two clock sources to be distributed on
up to 10 outputs as shown in Figure 7. The outputs can be
enabled and disabled by means of a serially programmable
register, which is also used to select the clock source.
Q0
ADN4670
Q0
NODE 0
11236-007
CLK
11236-006
CLK
Figure 7. ADN4670 Application Distributing a Clock Source to 10 Nodes via
Point-To-Point LVDS Connections
Any buffer adds a small amount of jitter when inserted between
the initial LVDS output and the eventual LVDS input, but the
ADN4670 has been designed to have low additive jitter of
<300 fs. Skew between the 10 outputs is kept to less than 30 ps
with clock signals of up to 1.1 GHz.
CLOCK DISTRIBUTION USING M-LVDS
Another option for clock distribution is using M-LVDS
transceivers to distribute the clock to up to 32 nodes in a multidrop (or multipoint) topology. Type 1 M-LVDS receivers (such
as in the ADN4690E to ADN4693E) are suited to such
applications because there is no offset in the receiver threshold
(this offset can result in duty cycle distortion for a clock signal).
The ADN4690E to ADN4693E M-LVDS transceivers with Type
1 receivers also have additional slew-rate limiting of the edges
from the driver outputs, which further limits radiated emissions
and the effect of reflections from stubs.
Rev. 0 | Page 4 of 12
Application Note
AN-1177
DIFFERENTIAL SIGNALLING AND LVDS/M-LVDS
100Ω
DIN1
DOUT1–
DOUT2+
ROUT1
RIN1–
RIN2+
100Ω
ROUT2
RIN2–
GND
M-LVDS
3V
M-LVDS
LVDS
2V
1V
0V
–1V
0V TO
2.4V
–1V TO
3.4V
1.05V
LVDS
250mV
450mV
MIN
VOD
MAX
VOD
480mV
650mV
MIN
VOD
MAX
VOD
Figure 10. LVDS and M-LVDS Signaling Levels
0.3V
|VOD|
VOD
0V
–0.3V
11236-008
(VOUT+ – VOUT–)
Figure 8. LVDS Output Levels
The differential voltage on an LVDS or M-LVDS bus is
generated by a driver current source. Noninverting LVDS driver
outputs or receiver inputs are generally denoted with a + and
inverting driver outputs or receiver inputs with a −.
Pin names are shown for the ADN4663 2-channel LVDS driver
and ADN4664 2-channel LVDS receiver in Figure 9. M-LVDS
follows the convention of RS-485 physical layer transceivers in
naming the bus lines A for the noninverting signal and B for the
inverting signal, or Y and Z for driver outputs on a full-duplex
transceiver.
M-LVDS has slew-rate limited drivers to enhance the robustness
of the signaling when there are additional impedance
discontinuities from multiple drivers/receivers and stubs. This
means that M-LVDS is limited to lower data rates compared to
LVDS. The ADN4690E through ADN4697E are available with
options for 100 Mbps or higher speed 200 Mbps. Another
characteristic of M-LVDS is increased driver strength, resulting
in a minimum output voltage swing |VOD| of 480 mV and a
maximum of 650 mV with a load of 50 Ω (two termination
resistors of 100 Ω, one either end of the bus).
RECEIVER THRESHOLDS
The receiver thresholds are the differential voltage levels above
or below which the received signal is considered a Logic 1 or a
Logic 0. For LVDS, a positive VOD >= +100 mV corresponds to
a Logic 1 and a negative VOD <= -100 mV corresponds to a
Logic 0.
Rev. 0 | Page 5 of 12
11236-010
4V
DIFFERENTIAL OUTPUT VOLTAGE
|VOD|
The distinction between LVDS and M-LVDS and other
differential signaling standards is that they have a low output
swing. The differential output voltage and common mode range
specifications of LVDS and M-LVDS are shown in Figure 10. For
LVDS, the output voltage swing, |VOD|, is a minimum of 250 mV
and a maximum of 450 mV with a load of 100 Ω. This allows
low power operation and ensures that while transitions are fast,
to allow high data rates, the reduced output swing means that
the slew rate is not too severe. Rise and fall times are generally
in the region of hundreds of picoseconds, resulting in slew rates
of around 0.5 V/ns to 2.5 V/ns.
COMMON-MODE VOLTAGE
1.35V
VOUT–
ADN4664
RIN1+
GND
LOGIC 1
VOC = 1.2V
DOUT1+
Figure 9. ADN4663 and ADN4664 2-Channel LVDS Point-to-Point
For LVDS and M-LVDS, one signal line is noninverting (that is,
high for a Logic 1 and low for a Logic 0) and the other signal
line is inverting (that is, the complement of the noninverting
signal). The difference in voltage between the two signal lines
is termed the differential voltage, VOD. VOD is also shorthand for
the magnitude of the differential voltage (positive or negative),
or |VOD|. The two signal lines each have a maximum voltage
swing of |VOD|, centered on the common-mode voltage, VOC
(also referred to as the offset voltage, VOS). The differential
voltage swings around 0 V. Typical LVDS signal levels are
shown in Figure 8, together with the differential signal VOD
and common-mode voltage VOC. In this figure, VOUT+ is the
noninverting signal and VOUT− is the inverting signal.
VOUT+
ADN4663
DOUT2–
DEFINITIONS AND OUTPUT LEVELS
LOGIC 0
VCC
DIN2
The high noise immunity arises because typically a noise
source couples equally onto both signal lines, leaving the
differential signal unaffected. Emissions from differential
signaling are low due to the tight coupling between the two
complementary signal lines when using a typical medium
(twisted pair cable, or closely placed strip line).
LOGIC 1
VCC
11236-009
Differential transmission is communication where two
complementary signals are transmitted, with the received signal
comprising the difference between the two signal lines. This
form of communication, used by both LVDS and M-LVDS,
has two distinct advantages, high noise immunity and low
emissions.
AN-1177
Application Note
For Type 1 M-LVDS receivers, a positive VOD ≥ +50 mV
corresponds to a Logic 1 and a negative VOD ≤ −50 mV
corresponds to a Logic 0.
LOGIC 1
M-LVDS TYPE 1
RECEIVER
OUTPUT
LOGIC 1
LVDS
LOGIC 1
800
600
400
0.15
0.10
UNDEFINED
M-LVDS
200
0.10
0
0
–0.05
–0.05
–0.10
–0.10
LOGIC 0
LOGIC 0
LOGIC 0
*LOGIC 1 FOR LVDS RECEIVERS WITH FAILSAFE
–0.15
Figure 11. Receiver Thresholds for LVDS and M-LVDS
With M-LVDS, any node on the bus can transmit, but when
no node is active, all driver outputs are disabled. As with LVDS,
this results in a differential output voltage in the undefined
region for Type 1 receivers. In order to provide a failsafe
condition, M-LVDS defines Type 2 receivers that have an
offset receiver threshold of >= +150 mV for a logic high and
<= +50 mV for a logic low. This means that the failsafe output
from Type 2 M-LVDS receivers is a logic low. Receiver
thresholds are shown in Figure 11 for LVDS receivers, M-LVDS
Type 1 receivers and M-LVDS Type 2 receivers.
TRANSMISSION DISTANCE
0
5
10
15
20
25
CABLE LENGTH (m)
Figure 12. Cable Length (Twisted-Pair) vs. Data Rate for Some Typical LVDS
and M-LVDS Applications
Other factors influencing the maximum distance include:
•
•
•
The transmitter specifications.
Other transmission medium components, such as vias (on
PCB traces) or connectors for cables.
For M-LVDS or multi-drop LVDS, the number of nodes on
the bus and the stub lengths.
TIA/EIA-644 (LVDS) and TIA/EIA-899 (M-LVDS) recommend
testing intended cable lengths in the application if possible, due
to the multiple factors involved that affect the possible cable
length. This allows the jitter on the received signal to be
measured, providing a guide as to how practical a given cable
type and length is. Measurements can be taken using an eye
diagram; the ADN4696E driver output is shown in Figure 13.
Both LVDS and M-LVDS transmission distances are affected by
two main factors: the transmission medium and the data rate.
The standard deciding factor of whether a given transmission
distance is practical, is how much jitter is observed by receiving
nodes. This is application dependent; some applications require
5% or less jitter, whereas others tolerate up to 20%.
PCB traces typically allow transmission distances on the order
of tens of centimeters, whereas twisted pair cable allows
transmission on the order of meters for LVDS or tens of meters
for M-LVDS. Different specifications of PCB construction or
cable types affect the signal differently and thus have an impact
on the maximum transmission distance.
Higher data rates greatly constrain the transmission distance;
LVDS at 1 Gbps might only be transmitted across high-quality
cables of 1 meter (possibly with additional signal conditioning),
but at 100 Mbps may be transmitted across 10 meters
Rev. 0 | Page 6 of 12
200mV/DIV
UNDEFINED
11236-012
0.05
UNDEFINED*
1ns/DIV
Figure 13. ADN4696E Driver Output Eye Diagram
11236-013
0.05
–0.15
1000
M-LVDS TYPE 2
RECEIVER
OUTPUT
0.15
0
1200
DATA RATE (Mbps)
LVDS
RECEIVER
OUTPUT
11236-011
DIFFERENTIAL INPUT VOLTAGE (VIA – VIB) [V]
In between these voltage thresholds is the transition region.
If an input signal remains at a voltage level between the
thresholds, the receiver output is undefined under LVDS; it
can be high or low. This can occur if no active LVDS driver is
connected to the receiver, or if there is a short circuit. Analog
Devices LVDS receivers incorporate a failsafe feature, so that in
these cases, the receiver output is high.
(depending on the cable type). M-LVDS can generally be
transmitted across longer cables due to the increased driver
strength, but data rates of hundreds of Mbps require shorter
cables than data rates of only tens of Mbps. Figure 12 provides
a general indication of the combinations of LVDS and M-LVDS
data rates and cable lengths typical for some applications.
Application Note
AN-1177
TERMINATION AND PCB LAYOUT
High speed communication links, such as those used for LVDS
and M-LVDS, should be considered in the context of transmission line theory, whether cables or PCB traces are used. The
high data rates of LVDS and M-LVDS require fast rise times,
meaning that impedance discontinuities and the end of the
communication link can significantly affect the transmitted
signal as it propagates from the driver to the far ends of the bus.
To avoid degradation of the signal, controlled impedances along
the communication medium, as well as proper termination, are
required.
Z0
DRIVER
R+
D–
Z0
R–
RECEIVER
Z0 = RT (TERMINATION MATCHES CABLE/TRACK IMPEDANCE)
Sharp turns or a series of bends in the PCB traces can also affect
the signal quality. Generally, turns in the PCB traces should be
minimized and kept to 45-degree angles (ideally with curves
rather than sharp angles).
Skew can be introduced between the two signals in a differential
pair if one signal follows a longer trace than the other does. It
may not always be possible to have traces exactly the same
length, but PCB layout should attempt to keep the trace lengths
matched.
Rx
RT
11236-014
D+
Tx
reducing susceptibility to common-mode noise. One difficulty
that arises is that if the traces need to move apart, for example,
to reach a connector, then a change in impedance between the
signals is introduced. It can be preferable to relax how closely
the signals are coupled, but maintain consistent spacing and
track thickness across the entire link.
Figure 14. Point-to-Point Termination
The termination resistor should match the impedance of the
communication medium; for LVDS, this is usually 100 Ω. For a
simple point-to-point link, it is only necessary to terminate the
end of the bus furthest from the driver, as shown in Figure 14.
For multi-drop buses, the same termination can be used if the
driver is at one end of the bus. Otherwise, both ends of the bus
need to be terminated.
Connectors should be chosen to minimize any difference in
impedance that they present on a bus, and cables or backplanes
should also match the impedance of PCB traces where possible.
Backplane connections can add significant capacitance to the
bus and it may be necessary to reduce the data rate or PCB
trace distances to allow for any degradation of the data signal
that occurs.
With M-LVDS, both ends of the bus are terminated, and the
drivers are designed with increased drive strength, partly to
accommodate the double termination (the effective load is
50 Ω rather than 100 Ω).
CONTROLLED IMPEDANCES
One difficulty in LVDS and M-LVDS links is providing a
consistent controlled impedance across the bus. For links across
a single PCB, impedance discontinuities can easily arise from
vias, mismatches in trace lengths between each signal in a
differential pair, and changes in the spacing between tracks, or
the size of tracks.
For differential signaling on a PCB, the two signal traces are
usually placed close together and tightly coupled. This means
that the signals have a common field, cancelling emissions and
11236-015
Some devices have built-in termination. This termination may
need to be disabled if the device is located at the wrong point
on the bus for termination, or if there is already proper
termination on the bus. If there are two or more 100 Ω resistors
for LVDS, or more than two for M-LVDS, then the bus is overterminated. This results in reduced signal amplitude and
increased reflections, combining to decrease noise immunity,
degrade timing accuracy and reducing the maximum
transmission distance.
Figure 15. EVAL-ADN469xEFDEBZ Customer Evaluation Board
An example high speed PCB layout for M-LVDS is shown
in Figure 15, the EVAL-ADN469xEFDEBZ evaluation board
for full-duplex ADN469xE family M-LVDS transceivers.
Track lengths on A, B, Y, and Z are matched and have a
50 Ω impedance created using a 4-layer board layout. The
termination resistor placement is next to the device pins. The
circuit does not fully correspond to an application layout
because there are additional components, such as test points
and jumper options.
Rev. 0 | Page 7 of 12
AN-1177
Application Note
JITTER, SKEW, DATA ENCODING, AND SYNCHRONIZATION
With high speed differential signaling, such as LVDS and
M-LVDS, accurate timing is critical to the performance of a
system. PCB traces, connectors, and cabling can degrade the
performance of data and clock signals, requiring that a margin
for error is also present in system timing. This means that
careful timing analysis may be required to achieve the maximum throughput on an LVDS or M-LVDS communication link.
Modern FPGAs and processors also have built-in capabilities to
correct for timing errors, although there may be clearly defined
limits to the amount of jitter tolerated, for example.
WHAT IS JITTER?
Jitter refers to the apparent movement of a signal edge with
respect to the ideal time position of that signal edge. If a
periodic signal is observed on an oscilloscope, the edges
literally jitter back and forth with respect to the reference point.
one type of deterministic jitter and refers to the time difference
between each cycle compared to the ideal. Periodic jitter is also
recorded as a peak-to-peak value, that is, the difference between
the longest and shortest periods observed
WHAT IS SKEW?
There are different definitions for skew, several of which are
typically considered in designing high speed LVDS links. The
most basic definition of skew is the difference in propagation
time between the two signals in a differential pair. This means
that edge transitions on one signal in a pair will not match up
exactly with transitions on the complementary signal (the
crossover will be asymmetric).
D–
INPUT
D+
IDEAL
D–
IDEAL
OUTPUT
tPLH = tPHL
TIE
D+
ACTUAL
(ONE
PASS)
tPLH
tPHL
D–
ACTUAL
OUTPUT
D+
EYE
PULSE SKEW
(tPHL – tPLH)
11236-016
JITTER
(PEAKTO-PEAK)
11236-017
ACTUAL
(MULTIPLE
PASSES)
Figure 17. Waveforms Illustrating Pulse Skew Calculation
Figure 16. Waveforms Showing Time Interval Error, Jitter and Eye
Jitter can be quantified simply as time interval error, the time
difference between when a signal edge occurs, and when it
should occur. Usually in order to determine the sources of
jitter, a large number of TIE samples are recorded to build a
histogram, from which deterministic jitter can be separated
from random jitter. Total jitter can be quantified as a peak-topeak value when bounded to a specific quantity of samples.
The peak-to-peak value means the time difference between
the earliest and latest edge observed during sampling.
Peak-to-peak jitter can be seen visually if multiple waveform
samples are overlaid on an oscilloscope display (infinite
persistence), as shown in Figure 16. The width of the overlaid
transitions is the peak-to-peak jitter, with the clear area inbetween referred to as the eye. This eye is the area available for
sampling by a receiver.
Random jitter occurs due to noise, both electrical and thermal.
The result is a Gaussian distribution to the time error, with this
error introduced as random jitter. The jitter is unbounded;
when more samples are recorded, the probability function
continues to grow.
Deterministic jitter is, by contrast, bounded. There is a fixed
amount of this jitter in the system due to specific factors, such
as the board layout and driver performance. Periodic jitter is
Pulse skew on a differential signal refers to the difference
between the low-to-high transition time (tPLH) and the high-tolow transition time (tPHL). This results in duty cycle distortion,
that is, the bit period is longer or shorter for a Logic 1 or Logic 0.
Pulse skew is illustrated in Figure 17. The blue waveform
corresponds to an input signal, the green waveform to an ideal
output (where propagation times on high-to-low and low-tohigh transitions are matched), and the red waveform to an
actual output, where the difference between tPLH and tPHL results
in pulse skew.
Channel-to-channel skew and part-to-part skew are some of
the most important parameters in typical LVDS applications
because they have multiple data channels that need to remain
synchronized. Channel-to-channel skew refers to the difference,
across all channels in a part, between the fastest and slowest
low-to-high transition, or the fastest and slowest high-to-low
transition (whichever is larger). Part-to-part skew extends this
concept to channels across multiple parts.
Skew across multiple channels (on one or multiple parts) is
illustrated in Figure 18. The blue waveform corresponds to an
input signal, with the four red waveforms comprising output
channels on one or more parts. The difference between the
fastest and slowest tPLH is calculated, along with the difference
between the fastest and slowest tPHL. The channel-to-channel or
Rev. 0 | Page 8 of 12
Application Note
AN-1177
part-to-part skew is the greater of these differences (in the case
of Figure 18, the difference between the fastest and slowest tPHL).
D–
INPUT
D+
period corresponding to one data bit (single data rate, SDR)
or two data bits (double data rate, DDR). For serial LVDS
transmission, a frame clock may also be transmitted. An
example of ADC source-synchronous LVDS outputs for SDR
and DDR is shown in Figure 19.
SAMPLE N
D–
ACTUAL
OUTPUT
ANALOG
INPUT
tPLH(FAST)
D+
SAMPLE N + 1
D–
ACTUAL
OUTPUT
(2ND)
SAMPLE N + 2
INTERNAL CLOCK:
tPHL(FAST)
CLK+
D+
CLK–
ACTUAL
OUTPUT
(3RD)
D–
LVDS OUTPUTS:
tPLH(SLOW)
DCO+
D+
DCO–
D–
D0+
tPHL(SLOW)
D0–
D+
CHANNEL-TO-CHANNEL
OR PART-TO-PART SKEW
(tPHL(SLOW) – tPHL(FAST)
> tPLH(SLOW) – tPLH(FAST) )
SAMPLE N – 6
BIT 0 (LSB)
SAMPLE N – 7
BIT 9 (MSB)
SAMPLE N – 6
BIT 9 (MSB)
SDR
(10 CHs)
tPHL(SLOW)
– tPHL(FAST)
D9+
D9–
11236-018
tPLH(SLOW)
– tPLH(FAST)
SAMPLE N – 7
BIT 0 (LSB)
D0/D5+
Figure 18. Waveforms Illustrating Channel-to-Channel or Part-to-Part Skew
Both channel-to-channel skew and part-to-part skew result in
parallel data channels received out of phase relative to each
other, even if they were synchronized at the transmitting end.
This can cause difficulties in sampling across multiple channels.
DATA ENCODING AND SYNCHRONIZATION
The challenges for LVDS timing stem not only from the
high speed transmission, but also from the data encoding.
In many LVDS applications, in order to increase bandwidth,
multiple parallel LVDS channels are used to transmit data.
The transmitter must synchronize data transmitted on these
channels and the receiver needs to sample each channel at the
appropriate point so that data can be received at the same time
across channels.
In LVDS applications using few channels, serial data is typically
transmitted and at higher speeds. The high speed requires the
receiving device to synchronize quickly with the incoming data
stream, and, in addition to accurately sampling each bit, the
receiving device needs to detect frames of data in the incoming
bit stream.
To help the receiving device synchronize with the received data,
a clock may be transmitted with the data channels. This is
described as source-synchronous data transmission. There are
several methods of transmitting the clock with the data. The
clock may be transmitted as a parallel channel, with the clock
D0/D5–
DDR
(5 CHs)
BIT 0
(LSB)
BIT 5
SAMPLE N – 7
D4/D9+
BIT 4
D4/D9–
BIT 9
(MSB)
BIT 0
(LSB)
BIT 5
SAMPLE N – 6
BIT 4
BIT 9
(MSB)
Figure 19. ADC input and Source-Synchronous LVDS Output Waveforms
An alternative to dedicated clock channels is to embed the clock
with the data. With the embedded clock method, fixed bits are
inserted into the data stream, allowing a receiving node to
detect these bits and synchronize with the incoming data.
Channel-to-channel and part-to-part skew can be compensated
for when received by modern FPGAs, using a scheme termed
dynamic phase adjustment (DPA). The FPGA generates
multiple phases of the received source-synchronous clock and
matches each data channel to the best clock phase for sampling.
If DPA is not available, then a strict timing budget must be
adhered to. There must be a time interval remaining after
transmitter channel-to-channel skew and the sampling time
are subtracted from the bit period. This interval is termed the
receiver skew margin. The transmitter channel-to-channel skew
includes the skew across channels due to the transmitting node,
the skew due to the medium and the clock skew relative to
the data.
Rev. 0 | Page 9 of 12
11236-019
ACTUAL
OUTPUT
(4TH)
AN-1177
Application Note
ISOLATION
The circuit shown in Figure 21 is an isolated LVDS Interface
Circuit from the Lab (CFTL), demonstrating complete isolation
of an LVDS interface (see the References section). The ADuM3442
provides digital isolation of the logic inputs to the ADN4663
LVDS driver and the logic outputs from the ADN4664 LVDS
receiver.
External interfaces can be isolated from logic circuits to
prevent unwanted current flow that may damage or degrade the
operation of electronic components. Galvanic isolation, shown
in Figure 20, allows information flow, but prevents current flow.
Complete isolation of data signals and power is possible using
iCoupler® digital isolation and isoPower® power isolation.
Together with provision of isolated power using the ADuM5000,
a number of challenges to isolating LVDS links in industrial and
instrumentation applications are met that include the following:
POINT B
INFORMATION FLOW
NO CURRENT FLOW
PROTECT HUMANS/
EQUIPMENT
•
IMPROVE SYSTEM
PERFORMANCE
ISOLATION
BARRIER
Isolation of the logic signals to/from the LVDS drivers/
receivers, ensuring standard LVDS communication on the
bus side of the circuit.
Highly integrated isolation using just two additional widebody SOIC devices, the ADuM3442 and ADuM5000, to
isolate the standard LVDS devices, the ADN4663 and
ADN4664.
Low power consumption compared to traditional isolation
(opto-couplers).
Multiple channels of isolation. This circuit demonstrates
quad-channel isolation (in this case, two transmit and two
receive channels).
High speed operation; the isolation can operate at up to
150 Mbps, facilitating basic LVDS speed requirements.
ELIMINATE GROUNDING
PROBLEMS
•
Figure 20. Galvanic Isolation Allows Information Flow While Preventing
Ground Current Flow
Applications of isolation for LVDS and M-LVDS are safety
isolation and/or functional isolation of board-to-board, backplane, and PCB communication links.
•
•
An example of safety isolation is a system with an M-LVDS
backplane where one or more plug-in cards are at risk from high
voltage transients. Isolating the M-LVDS interface ensures that
such fault conditions do not affect other circuits in the system.
An example of an application where functional isolation is
beneficial is measurement equipment. Isolating LVDS links, for
example, between an ADC and FPGA, can provide a floating
ground plane to boost the integrity of measurement data,
minimizing interference from the rest of the application.
•
The circuit shown in Figure 21 isolates a dual-channel LVDS
line driver and a dual-channel LVDS receiver. This allows
demonstration of two complete transmit and receive paths
on a single board.
ADuM5000
GND 3.3V
VDD1
OSC
ISO 3.3V
VCC
REC
ISO 3.3V
REG
VDD1
ADN4663
DIN1
VISO
DOUT1+
DOUT1–
VDD2
DIN2
ADuM3442
IN1
ISO 3.3V
VCC
IN2
ADN4664
ROUT1
OUT1
ROUT2
OUT2
DOUT2+
DOUT2–
LVDS
BUS
RIN1+
RIN1–
R1 100Ω
RIN2+
R2 100Ω
RIN2–
FPGA
ISOLATION
BARRIER
Figure 21. Isolated LVDS Interface Circuit (Simplified Schematic, All Connections Not Shown)
Rev. 0 | Page 10 of 12
11236-021
ISOLATOR
11236-020
POINT A
Application Note
AN-1177
REFERENCES
Chen, Boaxing. 2006. “iCoupler® Products with isoPower™ Technology: Signal and Power Transfer Across Isolation Barrier Using MicroTransformers,” Technical Article, (Analog Devices).
IEEE Standard 1596.3-1996, “IEEE Standard for Low-Voltage Differential Signals (LVDS) for Scalable Coherent Interface (SCI)”.
Marais, Hein. 2009. “RS-485/RS-422 Circuit Implementation Guide,” Application Note AN-960, Analog Devices, Inc.
TIA/EIA-485-A Standard, “Electrical Characteristics of Generators and Receivers for Use in Balanced Digital Multipoint Systems”.
TIA/EIA-644 Standard, “Electrical Characteristics of Low Voltage Differential Signaling (LVDS) Interface Circuits”.
TIA/EIA-899 Standard, “Electrical Characteristics of Multipoint-Low-Voltage Differential Signaling (M-LVDS) Interface Circuits for
Multipoint Data Interchange”.
Watterson, Conal. 2012. “Controller Area Network (CAN) Implementation Guide,” Application Note AN-1123, Analog Devices, Inc.
Watterson, Conal. 2012. Circuit Note CN-0256, “Isolated LVDS Interface Circuit,” (Analog Devices, Inc.
RELATED LINKS
Resource
LVDS/M-LVDS web page
M-LVDS web page
CN-0256
AN-960
Description
Links to product pages and resources for LVDS drivers, LVDS receivers and M-LVDS transceivers
Introduction to and resources for the ADN4690E to ADN4697E family of M-LVDS transceivers
Circuit Note for Isolated LVDS Interface Circuit
Application Note for RS-485/RS-422 Circuit Implementation Guide
Rev. 0 | Page 11 of 12
AN-1177
Application Note
NOTES
©2013 Analog Devices, Inc. All rights reserved. Trademarks and
registered trademarks are the property of their respective owners.
AN11236-0-3/13(0)
Rev. 0 | Page 12 of 12